Artificial Intelligence and the Demand for Hours Worked

Abstract: This paper develops a theoretical model of how artificial intelligence (AI) affects the firm's labor demand decision on the intensive margin---specifically, the choice of hours worked per employee. We model AI as a new input that enters both the production function and the effective-hours technology. We find that worker hours increase with AI exposure if and only if the elasticity of AI's indirect impact on effective hours is greater than the elasticity of its direct impact. These theoretical results are tested with firm and household data from China.